| Since ancient times,handwritten Chinese characters is a habit that the Chinese nation has formed,persisted,and passed down.The input method of keyboard and mouse solves the problem of computer's Chinese character input,but the traditional writing habits are destroyed,it is not conducive to the inheritance of Chinese character culture.with the aid of tablet,touch screen and other auxiliary equipment input method needs additional professional equipment,and the price is expensive,not widespread.So this paper aims to study a kind of original handwriting extraction and Chinese character recognition method based on computer vision and machine learning,and realize a low cost,return to nature but adapt to the modern digital computer input mode.The main work and contributions of this article are summarized below.This paper presented an original handwriting extraction algorithm.The algorithm extracts written handwriting from the frame images captured by the camera in real time,and performs display and digital storage.Firstly,the skin color detection algorithm is used to acquire the position information of the grip finger,and the handwriting extraction target region is determined according to the spatial position relationship.Next,the coarse to fine strategy is used to eliminate noise and extract stroke information in each frame.The specific steps are as follows:using edge detection,"pre-segmenting" the image to obtain the position information of the area where there may be a stroke,thereby reducing the amount of computation.Using the special feature extraction operator for stroke features defined,the single-frame image was extracted from the pixels of a suspected stroke.Using continuous multi-frame images to eliminate interference noise and select true growth strokes in suspected strokes.Finally,according to the stroke information,the original handwriting was displayed in real time,and the handwriting was digitally stored and displayed in the form of pixel coordinate sequences.Based on Hidden Markov model,an online handwritten Chinese character recognition algorithm was proposed and implemented.Firstly,the inflection point algorithm is used to extract the inflection point of strokes,and the strokes are segmented into stroke segments.Secondly,the SVM method was used to classify the strokes to obtain the stroke coding sequence of the Chinese characters.Finally,the model was established,trained and recognized by the hidden Markov model.In this paper,40 common Chinese characters have been trained and identified,and the correct rate reaches over 81%.This paper used VS2013 development environment,designed and developed a set of original handwriting extraction and Chinese character recognition system software with MFC and OpenCV library function.With the above original handwriting extraction algorithm and handwritten Chinese character recognition algorithm,the system realizes the basic functions of the original handwriting,such as real time display and preservation,Chinese character recognition,as well as the additional functions of the original handwriting color setting,the clear screen,the full screen teaching mode et al..It can be demonstrated in the classroom teaching.The original handwriting extraction and Chinese character recognition method based on computer vision realized in this paper adopts an ordinary USB camera.The writing person naturally writes on plain white paper,which satisfies the traditional writing styles and habits,and ensures a smooth writing experience.It can be applied to digitization.Classroom teaching,personalized signatures,daily notes et al..The algorithm can also be transplanted to smart phones to solve the difficult problem of small screens and difficult input of texts,which has wide application value. |
